Skip to main content
. 2020 Jan 15;122(6):847–856. doi: 10.1038/s41416-019-0718-9

Fig. 4. Model predictions presented as logistic regression outcomes for multivariate and raw outcome for univariate analyses by time.

Fig. 4

Time is years to diagnosis for cases and years since final sample for controls. Model cut-offs giving 90.3% specificity are indicated by horizontal dashed lines. Grey lines represent individuals and circles the time the sample was taken. The thick line is a Loess fit over all points with 95% CIs in grey shading. Vertical dashed lines indicate the timepoint at which the Loess curve intersects the threshold for model CA125{5}.